Help, shared GI Both mode wierdness
Help, shared GI Both mode wierdness
How does shared GI Both mode suppose to work? I've did some simple test, but it seems that even with 'Both', Kray does not calculate the missing GI data not visible in previous frame, is this a bug?
- Attachments
-
- 'Both' mode doesn't seems to work
- Kray_BothMode.jpg (118.26 KiB) Viewed 4766 times
You use "precomputed" that only will show you the photonmap/celldistribution.
Since kray wount fire anymore photons than those found in first frame nothing can be updated. If you switch to "Photonmapping" instead and activate "Both" Kray will keep on fireing more and more FG rays to fill out the gaps in that porly irradiancemap. Raise the numbers of photons so you have a nice distribution over the whole object and then kray wount need to fire as much FG to fill in the gaps.
Since kray wount fire anymore photons than those found in first frame nothing can be updated. If you switch to "Photonmapping" instead and activate "Both" Kray will keep on fireing more and more FG rays to fill out the gaps in that porly irradiancemap. Raise the numbers of photons so you have a nice distribution over the whole object and then kray wount need to fire as much FG to fill in the gaps.
In "Photonmapping" and "Both" activated, will Kray be firing more photons or just FG? But how do I have good photon distribution over entire object (even those non in frame) with "Lightmap" mode?Silverlw wrote: If you switch to "Photonmapping" instead and activate "Both" Kray will keep on fireing more and more FG rays to fill out the gaps in that porly irradiancemap. Raise the numbers of photons so you have a nice distribution over the whole object and then kray wount need to fire as much FG to fill in the gaps.
ok so you use lightmaps. Set up ypur animation in lw's rendermenue so it all frameparameters are correct . for example framestep1 startframe 1 endframe 500. Clear you cached file from hd totally! take it away. Now press render sequence. If all goes well you should see lw stepping from frame 1 to 500 before it starts to render all. This is since kray will then fire all photons needed for the whole animation before actuall rendering starts. After that initial procedure kray will render with the already saved photons and update it's FG accordingly.
This technique is only needed for kray when you use lmaps since lmaps are traced via the camera (what it see's). Therefore there wount be any photons at all if you would switch camera view. With photonmapping it works different since photons are traced from the lightsources and shoot everywhere from the beginning.
This technique is only needed for kray when you use lmaps since lmaps are traced via the camera (what it see's). Therefore there wount be any photons at all if you would switch camera view. With photonmapping it works different since photons are traced from the lightsources and shoot everywhere from the beginning.
Great! Thanks a lot for the explaination, it works as you said.
one thing i noticed with lightmap shared GI and sequence, the part where the camera stays longer will gather more photons than parts that just pan by quickly.
so, is it better to rig for photonmap rather than lightmap if it were to be shared GI and camera movement? what are their(lightmap vs photonmap) pros and cons, besides lmap being able to use backdrop to emit photons, view dependant, works great for large pane of light, bad for light bulbs(tiny lightsources)?
Thanks again for the quick reply.
one thing i noticed with lightmap shared GI and sequence, the part where the camera stays longer will gather more photons than parts that just pan by quickly.
so, is it better to rig for photonmap rather than lightmap if it were to be shared GI and camera movement? what are their(lightmap vs photonmap) pros and cons, besides lmap being able to use backdrop to emit photons, view dependant, works great for large pane of light, bad for light bulbs(tiny lightsources)?
Thanks again for the quick reply.
Well when doing animation pros and cons of LM vs. PM are also that if you had to render an animation with camera going through separate rooms you will need to shoot huge amount of photons in Photonmap mode. This leads to having places with a lot more photons in some rooms and alot less in others which then causes bleeding over the thin walls.tajino wrote:Great! Thanks a lot for the explaination, it works as you said.
one thing i noticed with lightmap shared GI and sequence, the part where the camera stays longer will gather more photons than parts that just pan by quickly.
so, is it better to rig for photonmap rather than lightmap if it were to be shared GI and camera movement? what are their(lightmap vs photonmap) pros and cons, besides lmap being able to use backdrop to emit photons, view dependant, works great for large pane of light, bad for light bulbs(tiny light sources)?
Thanks again for the quick reply.
With LM you avoid such a problem because photons will be fired wherever camera goes and photon density will be more or less constant.
Take a look at this vray tutorial - it works very similar to Kray LM.
http://www.spot3d.com/vray/help/150R1/t ... _imap2.htm
- Jure
Baking and UV packing issues
Jure, thanks for the clarification.
One more question, is it possible for current version of kray to have objects flagged to be unseen by photon and fg rays but seen by camera and reflection/refraction rays? It would be great that way to be able to mix trees and environment that has precalculated baked map on, and could save precious photons to be fired/gathered on them? I think this will be good for scenes with exterior lush vegitation and hardscape/architecture.
The only way i can think of other than comping is to bake the buildings or whatever neccesary, reapply bakemap and render them with plain old lw or fprime.
:::: In need of optimized, decent and fast UV packer ::::
But the problem is making an optimized atlas uv map is always a tedious task, having the lack of good uv packing tools in modeler except the recent PLGuv tools that does a pretty good packing job but is extremely slow with polycount above 5k. other solution that i've tried was with DeepUV, has a very speedy L-packer for UV, but leaves a lot of gaps.
These days no one seems to be interested in baking and uv packing anymore as i see it. Even new software like modo doesn't have a uv packer. Max has got a great one, but bringing meshes in and out of it screws n-gons. Others might have better ones i think.
Anyone has a better solution/ideas to share here? Thanks.
One more question, is it possible for current version of kray to have objects flagged to be unseen by photon and fg rays but seen by camera and reflection/refraction rays? It would be great that way to be able to mix trees and environment that has precalculated baked map on, and could save precious photons to be fired/gathered on them? I think this will be good for scenes with exterior lush vegitation and hardscape/architecture.
The only way i can think of other than comping is to bake the buildings or whatever neccesary, reapply bakemap and render them with plain old lw or fprime.
:::: In need of optimized, decent and fast UV packer ::::
But the problem is making an optimized atlas uv map is always a tedious task, having the lack of good uv packing tools in modeler except the recent PLGuv tools that does a pretty good packing job but is extremely slow with polycount above 5k. other solution that i've tried was with DeepUV, has a very speedy L-packer for UV, but leaves a lot of gaps.
These days no one seems to be interested in baking and uv packing anymore as i see it. Even new software like modo doesn't have a uv packer. Max has got a great one, but bringing meshes in and out of it screws n-gons. Others might have better ones i think.
Anyone has a better solution/ideas to share here? Thanks.